Reviewer requests for power calculations for completed studies

Although the reviewer has asked for power calculations, we do not believe that these would be helpful, and they may in fact promote misinterpretation of our results. We believe that the irrelevance of power for interpreting completed studies is well established, and some references are:

Altman DG, Schulz KF, Moher D, Egger M, Davidoff F, Elbourne D, Gøtzsche PC, Lang T, for the CONSORT Group. The revised CONSORT statement for reporting randomized trials: explanation and elaboration. Ann Intern Med 2001; 134:663-94. Page 670: "There is little merit in calculating the statistical power once the results of the trial are known"

Cox DR. Planning of Experiments. New York: Wiley, 1958: page 161. "Power . . . is quite irrelevant in the actual analysis of data."

Tukey JW. Tightening the clinical trial. Controlled Clinical Trials 1993; 14:266-285. Page 281: "power calculations ... are essentially meaningless once the experiment has been done."

Goodman SN, Berlin JA. The use of predicted confidence intervals when planning experiments and the misuse of power when interpreting results. Ann Intern Med 1994; 121:200-6.

Hoenig JM, Heisey DM. The abuse of power: the pervasive fallacy of power calculations for data analysis. American Statistician 2001; 55:19-34.

Senn, SJ. Power is indeed irrelevant in interpreting completed studies. BMJ 2002; 325: 1304.

The reason for this consensus is that reasoning via p-values and power is indirect and unreliable, while estimates and confidence intervals show more simply and directly the strength of any evidence against important effects. We have provided estimates and confidence intervals for all of our main results.

We understand that reporting of sample size calculations is common, and we are aware that the CONSORT guidelines for reporting randomized clinical trials do call for reporting a priori power calculations if those were the basis for choosing the target sample size. This reporting, however, is for the purpose of disclosing the target sample size and primary outcome, rather than for any direct role in the interpretation of the results. Unfortunately, including power calculations tends to promote readers' misinterpretation of any result with p>0.05 as proving that there is no effect (see P-value Fallacy). We have therefore instead directly stated all relevant aspects of our study design, including the target sample size, rather than indirectly disclosing this information within a power calculation.